JP2005165984A - Method, system and program for detecting vertex of human face - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a new method, system and program for detecting a top of head in a robust way by detecting the boundary between the top of head and its background with precision and high speed. <P>SOLUTION: The method for detecting the top of head of the human face from within an image containing the human face includes detecting the human face, setting a crown detection window at the top of the human face, calculating the feature quantity of the image within the top of head detection window vertically, and setting a position where the feature quantity varies beyond a threshold as the top of head of the human face. Thus, the top of head of the human face can be detected with precision and high speed. <P>COPYRIGHT: (C)2005,JPO&NCIPI

Description

  The present invention relates to pattern recognition and object recognition technology, and in particular, detection of the top of a face in a face image for accurately detecting the top of the person's face from a face image in which the face of the person is reflected. The present invention relates to a method, a parietal detection system, and a parietal detection program.

  With recent advances in pattern recognition technology and information processing devices such as computers, the recognition accuracy of characters and voices has improved dramatically, but images that show people, objects, and scenery, such as digital Of the pattern recognition of images captured by still cameras, etc., it is still known that it is still a very difficult task, especially in terms of accurately and quickly identifying whether a human face is reflected in the image. ing.

  However, automatic identification of whether or not a person's face is reflected in the image, and who the person is, by using a computer or the like, establishes biometric recognition technology and improves security. It has become an extremely important theme for speeding up crime investigations, organizing image data and speeding up search operations, and many proposals have been made on such themes.

  For example, in the following Patent Document 1 or the like, for a certain input image, first, the presence / absence of a human skin color area is determined, a mosaic size is automatically determined for the human skin color area, the candidate areas are mosaicked, and the human face dictionary By calculating the distance to the image, the presence or absence of a human face is determined, and the human face is cut out to reduce false extraction due to the influence of the background, etc., and to efficiently find the human face in the image efficiently I have to.

Further, in the following Patent Document 2, etc., feature point extraction of a face image used for distinguishing each individual or group (for example, racial group) is automatically performed at high speed and simply by using a predetermined algorithm. Like to do.
Japanese Patent Laid-Open No. 9-50528 JP-A-8-77334

By the way, in many cases, a face photograph (face image) of a person indispensable for a passport, an identification card, or the like is finely set in size, direction, size, position, etc. of the person's face.
For example, it goes without saying that there is no background and that you do not wear accessories such as a hat, and that the person's face is facing the front, that the person's face is in the center of the photo, The position of the top of the face being at a certain distance from the upper frame of the photo, etc. are stipulated in detail, and as a rule, photos (face images) that deviate from the standard are not adopted.

However, the person's face is not facing the front or wearing an accessory such as a hat, but the size and position of the face is simply shifted again, It is unreasonable to have to re-shoot, and there is a problem in that it requires significant effort and cost to the user.
For this reason, methods for solving the above-described problems using digital image processing technology, which is a technological field that has been remarkably developed in recent years, have been studied.

  For example, a necessary person's face image is directly acquired as digital image data by a digital still camera using an electronic image sensor such as a CCD or CMOS, or an analog photograph (silver salt photograph) in which a person's face is photographed in advance. The digital image data is obtained as digital image data using an electro-optical image reading device such as a scanner, and this digital image data is obtained by using an image processing system including a general-purpose computer such as a PC and general-purpose software. It is considered to solve the above problem by performing simple image processing such as enlargement, reduction, and movement of the face image as appropriate without impairing the image quality.

On the other hand, if the number of images to be processed is small, the processing operation can be performed directly by a human using a general-purpose input / output device such as a mouse, a keyboard, or a monitor. Is enormous, it becomes necessary to perform the processing automatically using the conventional technique as described above.
However, in order to realize automation of human face image processing in this way, it is necessary to accurately recognize the contour of the face, particularly the position of the top of the person's face. Generally, it does not always have a clear boundary depending on the hair style, lighting conditions at the time of shooting, and other conditions, so it is often impossible to read clearly with a conventional edge detection filter, etc., so that part is automatically recognized. It ’s difficult.

  Therefore, the present invention has been devised in order to effectively solve such problems, and its purpose is to accurately and rapidly detect the top of a human face that is difficult to recognize automatically by edge detection. The present invention provides a novel method for detecting the top of the head of a face image, a system for detecting the top of the head, and a program for detecting the top of the head, which can detect the top of a human face that is robust.

[Invention 1]
In order to solve the above-described problem, a method for detecting the top of a person's face image according to the first aspect of the present invention includes:
A method for detecting a top of a person's face from an image including a person's face, wherein a face image in a range not including the top of the person's face is detected, and the person is positioned above the detected face image. After setting the top detection window having a size that includes the top of the face, the image feature amount in the top detection window is calculated in the vertical direction, and the position where the image feature amount changes to a threshold value or more is calculated. It is characterized by the fact that it is the top of the face.

As described above, the present invention uses the face detection frame to select a region that is very likely to include the top of the person's face, and after setting a top detection window for that portion, Image feature amount is obtained, and the top of the human face is detected based on the change in the image feature amount.
In other words, since the image feature amount of the top of the person's face and the background generally changes greatly, the change in the image feature in the top detection window is obtained to obtain the top detection window. It is possible to easily and reliably detect the top of the heel included therein.

[Invention 2]
The method of detecting the top of the human face of the invention 2 includes:
In the head top detection method according to the first aspect of the invention, the magnitude of luminance is used as the image feature amount.
That is, the face image for the ID photo is in a capless state, and its head is mostly covered with dark hair such as black or maroon. It can be considered that the luminance value of the image and the luminance value of the background are greatly different. That is, the luminance changes abruptly at the boundary between the background and the top of the head.

Therefore, by using the luminance of the pixel that is remarkably different depending on the region as an image feature amount for distinguishing the top from the background, it is possible to easily and reliably detect the top of the human face. Become.
[Invention 3]
The method of detecting the top of the human face of the invention 3
In the head top detection method according to the first aspect of the present invention, a hue angle is used as the image feature amount.

  That is, the above invention 2 is made on the assumption that the hair color of a person is dark, such as black or maroon, but the color difference or change of the hair due to race or aging, or dyeing. Depending on the hair, the color is not only dark but also any color that does not occur naturally. Further, in some hairs of a color other than dark, the luminance is almost the same as the background (for example, blond hair), and there is a case where a large change cannot be obtained in the image feature amount based on the luminance. .

For this reason, in the present invention, the magnitude of the hue angle is used as the image feature amount, and even if the luminance of the background does not show a large difference in the luminance of the top, the hue angle is thus determined. By obtaining the change in the image feature amount due to the difference in the head, the top of the human face can be detected easily and reliably.
[Invention 4]
The method of detecting the top of the human face of the invention 4 includes:
In the parietal detection method according to any one of the first to third aspects, the image feature value is an average value in a horizontal direction within the parietal detection window.

As a result, even if there is a case where the feature quantity of a part of a pixel is significantly different from the feature quantity of a neighboring pixel due to noise on the image or contamination of the original image, the feature quantity suddenly changes in that part. Since it is possible to avoid the change, such a noise portion or the like is not erroneously detected as the top of the head, and accurate head top detection can be performed.
[Invention 5]
The method of detecting the top of the human face of the invention 5 includes:
In the method for detecting the top of the head according to any one of the first to fourth aspects, the image feature amount is automatically switched according to an image using either or both of luminance and hue angle. .

As a result, similar to the fourth aspect, it is possible to avoid an abrupt change in the image feature amount due to noise on the image, contamination of the original image, or the like, and perform accurate head detection.
[Invention 6]
The method of detecting the top of the human face of the invention 6
In the method for detecting the top of the head according to any one of the first to fifth aspects of the present invention, the top detection window has a rectangular shape and has a width and height wider than the face width of the human face. It is characterized by.

Accordingly, the top of the person's face to be detected can be reliably captured in the top of the head detection window, so that the top of the head can be reliably detected.
[Invention 7]
The human face top detection system of the invention 7
A system for detecting the top of a person's face from an image including a person's face, the image reading means for reading the image including the person's face, and the head of the person's face from the image read by the image reading means Face detection means for detecting a face image in a range not including the top and setting a face detection frame in the detected range, and detection of the top of the head that includes the top of the human face above the face detection frame A vertex detection window setting means for setting a window, an image feature quantity calculation means for calculating an image feature quantity in the vertex detection window in the vertical direction, and an image feature quantity calculated by the image feature quantity calculation means is a threshold value. And a head top detection means for setting the position changing as described above to the head top of the human face.

As a result, the top of the head of the human face can be detected accurately and at high speed as in the first aspect of the present invention, and the top of the head can be detected by realizing each of these means by using dedicated hardware or a computer system. It becomes possible to automate.
[Invention 8]
The head part detection system in the human face image of the invention 8 is,
In the human face head top detection system according to a seventh aspect of the invention, the image feature amount calculation means calculates a luminance calculation unit that calculates the luminance of each pixel as an image feature amount, and calculates the hue angle of each pixel as an image feature amount. A hue angle calculation unit and an image feature amount selection unit that selects any one of these image feature amounts are provided.

That is, as shown in the inventions 2 and 3, since the image feature amount may not be accurately grasped only by the luminance or only by the hue angle, in the present invention, means for calculating the luminance and the hue angle respectively. The image feature quantity selection unit selects an optimum one of the image features calculated by any one of these means.
As a result, regardless of the hair color of the person's face, it is possible to always calculate the optimal image feature amount and accurately detect the top of the head.

[Invention 9]
The human face top detection system of the invention 9
In the human face head top detection system according to the seventh or eighth aspect, the head head detection means includes a threshold value setting unit for setting a threshold value of the image feature value, and a horizontal image feature value in the head head detection window. One or both of an average value calculation unit for obtaining the average of the above and the like are provided.

As a result, even if the feature amount of a part of a pixel is significantly different from the feature amount of a neighboring pixel due to noise on the image or contamination of the original image as in the fourth aspect, Since it is possible to avoid a sudden change in the feature amount, such a noise portion or the like is not erroneously detected as the top of the head, and accurate top detection can be performed.
[Invention 10]
The vertex detection program in the human face image of the invention 10 is
A program for detecting the top of a person's face from an image including a person's face, the image reading means for reading the image including the person's face, and the head of the person's face from the image read by the image reading means Face detection means for detecting a face image in a range not including the top and setting a face detection frame in the detected range, and detection of the top of the head that includes the top of the human face above the face detection frame A vertex detection window setting means for setting a window, an image feature quantity calculation means for calculating an image feature quantity in the vertex detection window in the vertical direction, and an image feature quantity calculated by the image feature quantity calculation means is a threshold value. It is characterized in that a computer is realized with a head-top detection means that uses the position changing as described above as the head of the person's face.

As a result, similarly to the first aspect, the top of the heel included in the top detection window can be detected easily and reliably.
In addition, since each of these means can be realized on software using a general-purpose computer system such as a personal computer (PC), it can be realized more easily and economically than a method of manufacturing and realizing a dedicated device. be able to. Also, in most cases, it is possible to easily achieve version upgrades such as function changes and improvements only by rewriting the program.

[Invention 11]
According to the eleventh aspect of the present invention, there is provided a program for detecting the top of a human face
In the human face head top detection program according to the tenth aspect of the present invention, the image feature amount calculating means includes:
A luminance calculation function for calculating the luminance of each pixel as an image feature quantity, a hue angle calculation function for calculating the hue angle of each pixel as an image feature quantity, and an image feature quantity selection function for selecting one of these image feature quantities; It is characterized by being able to demonstrate.

As a result, it is possible to calculate the optimal image feature amount and always accurately detect the top of the head regardless of the color of the human face hair, as in the case of the invention 8, and to achieve the same effects as the inventions 1 and 9. Can be obtained.
[Invention 12]
The vertex detection program in the human face image of the invention 12 is
In the human face head top detection program according to the tenth or eleventh aspect, the head head detection means includes a threshold value setting unit for setting a threshold value of the image feature value, and a horizontal image feature value in the head head detection window. One or both of an average value calculation unit for obtaining the average of the above and the like are provided.

  As a result, even if there is noise on the image or contamination of the original image as in the fourth aspect, such a noise portion or the like is not erroneously detected as the top of the head and accurate head detection is performed. Can do.

Hereinafter, the best mode for carrying out the present invention will be described in detail with reference to the accompanying drawings.
FIG. 1 shows an embodiment of a human face top detection system 100 according to the present invention.
As shown in the figure, the top detection system 100 includes an image reading unit 10 that reads an image including a human face, such as a face image for an ID photo, as described above, and an image read by the image reading unit 10. A face detection means 12 for detecting a human face from the inside and setting a face detection frame at that portion, and a top detection window W having a size including the top of the human face is set above the face detection frame. On the basis of the change in the image feature amount calculated by the image feature amount calculation means 16, the image feature amount calculation means 16 that calculates the image feature amount in the crown detection window W It is mainly comprised from the top part detection means 18 which detects the top part of the said person's face.

  First, the image reading means 10 specifies a visual person attached to a public ID such as a passport or a driver's license, or a private ID such as an employee ID, a student ID, or a membership card. A proof face photo for use, that is, a face image G without a cap and without a background, which includes the face of the person in the front direction only, is charged with a CCD (Charge Coupled Device) or a CMOS (Complementary Metal Oxide). A function of obtaining digital image data composed of pixel data of R (red), G (green), and B (blue) by using an imaging sensor such as a semiconductor) is provided.

Specifically, it is a CCD such as a digital still camera or a digital video camera, a CMOS camera, a vidicon camera, an image scanner, a drum scanner, or the like. A function of sequentially sending digital image data to the face detecting means 20 is provided.
The image reading means 10 has a data storage function, and the read face image data can be appropriately stored in a storage device such as a hard disk drive (HDD) or a storage medium such as a DVD-ROM. Yes. Further, when a face image is supplied as digital image data via a network or a storage medium, the image reading means 10 is not necessary or functions as a communication means or an interface (I / F). become.

Next, the face detection unit 12 detects a human face from the face image G read by the image reading unit 10 and sets a face detection frame F in the portion.
As will be described later, the face detection frame F has a rectangular shape that is smaller than the area of the front face of the human face, includes at least both eyes and lips around the nose of the human face, and the top of the human face. The size (area) does not include.

Note that the human face detection algorithm by the face detection unit 12 is not particularly limited, and for example, a conventional method as shown in the following document or the like can be used.
H. A. Rowley, S.M. Baluja, and T.R. Kanade,
“Neural network-based face detection”
IEEE Transactions on Pattern Analysis
and Machine Intelligence,
vol. 20, no. 1, pp. 23-38, 1998
According to this technology, a face image of an area including both eyes and lips of a human face but not the top of the head is created, a neural network is trained using this image, and a human face is detected using the trained neural network. To do. According to this disclosed technique, an area from both eyes to the lips is detected as a face image area.

  Further, the size of the face detection frame F is not invariant, and is appropriately increased or decreased depending on the size of the target face image. The head detection window setting unit 14 provides a function of setting a head detection window W having a size that includes the head of the human face above the face detection frame F set by the face detection unit 12. It has become. That is, a target region for accurately detecting the top of the person's face by subsequent means is selected from the face image G using the top detection window W.

The image feature quantity calculating means 16 further includes a brightness calculation section 22 that calculates the brightness of each pixel as an image feature quantity, a hue angle calculation section 24 that calculates the hue angle of each pixel as an image feature quantity, and changes in the feature quantities. The image feature quantity selection unit 26 selects one of these image feature quantities according to the above.
That is, as will be described in detail later, the image feature amount calculating unit 16 includes an image feature amount based on the luminance calculated by the luminance calculation unit 22, and an image feature amount based on the hue angle calculated by the hue angle calculation unit 24. The image feature amount selection unit 26 selects the most suitable one of them, and sends information about the selected one of the image feature amounts to the top portion detection means 18.

  The vertex detection means 18 further includes a threshold value setting unit 32 that sets a threshold value of the image feature value, and an average value calculation unit 34 that calculates the average of the image feature values in the horizontal direction in the vertex detection window W. The head and the background are separated based on the threshold set by the threshold setting unit 32, and the average value calculating unit 34 calculates the average of the horizontal image feature amounts of the top detection window W. The error of the local image feature amount due to noise or dirt is prevented.

In addition, each means 10, 12, 14, 16, 18, 22, 24, 26, etc. constituting the crown detection system 100 is actually composed of hardware such as a CPU and a RAM, as shown in FIG. It is realized by a computer system such as a personal computer (PC) comprising a dedicated computer program (software).
That is, as shown in FIG. 2, for example, a hardware for realizing the crown detection system 100 includes a CPU (Central Processing Unit) 40 that is a central processing unit that performs various types of control and arithmetic processing, and a main storage device. RAM (Random Access Memory) 41 used for (Main Storage), ROM (Read Only Memory) 42 which is a read-only storage device, and auxiliary storage devices (Secondary Storage) such as hard disk drive devices (HDD) and semiconductor memories 43, and an output device 44 including a monitor (LCD (liquid crystal display) or CRT (cathode ray tube)), an image scanner, a keyboard, a mouse, a CCD (Charge Coupled Device), and a CM. Between the input device 45 including an imaging sensor such as an OS (Complementary Metal Oxide Semiconductor) and the input / output interface (I / F) 46 and the like, a peripheral component interconnect (PCI) bus and an ISA (Industrial Standard); These are connected by various internal and external buses 47 such as a processor bus, a memory bus, a system bus, an input / output bus, and the like.

  For example, various control programs and data supplied via a storage medium such as a CD-ROM, a DVD-ROM, a flexible disk (FD), or a communication network (LAN, WAN, Internet, etc.) N are stored in an auxiliary storage device. And the program and data are loaded into the main storage device 94 as necessary, and the CPU 40 performs various control and calculation processes using various resources according to the program loaded into the main storage device 91. The processing result (processing data) is output to the output device 44 via the bus 47 and displayed, and the data is appropriately stored and saved (updated) in a database formed by the auxiliary storage device 43 as necessary. It is what you do.

Next, an example of a method for detecting the top of the head using the top of the head detecting system 100 having such a configuration will be described with reference to FIGS.
FIG. 3 is a flowchart showing an example of the method of detecting the top of the face image G that is actually a detection target.
As shown in the figure, first, after detecting the face included in the face image G from the face image G to be detected by the image reading means 10 in advance in the first step S101, the face detecting means 12 previously described. A face detection frame F is set on the detected human face.

For example, as shown in FIG. 4, the face image G to be detected as the top of the head of the present invention is limited to that which shows one person's face. The position is specified, and then a rectangular face detection frame F is set in the human face area as shown in FIG.
In the case of the face detection frame F shown here, it is a rectangular shape smaller than the area of the human face and has a size (area) that does not include the top of the human face. As long as the detection frame F does not include the top of the person's face, it is not always necessary to stick to the size and shape as illustrated. Each face image G shown in FIGS. 4 to 7A is within the standard in terms of the size of the captured human face and the horizontal position of the display frame Y, but the position of the top of the human face. Shows a state where the value is too low to reach the standard position.

Next, when the face detection frame F is set at the center of the human face in this way, the process proceeds to step S103, and the upper end of the face detection frame F is detected by the head detection window setting means 14 as shown in FIG. A rectangular top detection window W is set to the region of the top of the person's face.
Here, the size and shape of the parietal detection window W are not strict, and the size and shape include the parietal portion that is always the boundary between the background and the head above the forehead of the person's face. If there is, it is not particularly limited, but if it is too large, it will take a lot of time to detect the subsequent edges, etc. Conversely, if it is too small, the top of the head to be detected may not be included. is there. Therefore, for example, if a rectangular shape having a width and height wider than the head width of the human face is used as shown in FIG. It is thought that it can be done. In the example of FIG. 6, the crown detection window W is set in close contact with the upper end of the face detection frame F. However, the crown detection window W does not necessarily need to be in close contact with the face detection frame F. In short, it suffices that the top detection window W keeps a predetermined positional relationship with respect to the face detection frame F.

Next, if the top detection window W is set for the target image in this way, the process proceeds to the next step S105, where the brightness of each pixel in the top detection window W is calculated, and the brightness value is calculated. The average value in the horizontal direction (X direction in FIG. 6) is calculated for each line.
For example, the number of pixels in the crown detection window W is “150 (pixel)” in the horizontal direction (X direction in FIG. 6) × vertical direction (Y direction in FIG. 6) “100 (pixel)” = “15000 (pixel). ) ", The luminance values of all the pixels for each line are summed and divided by 150 to obtain the average value of the luminance values for each line. The image feature amount for each line.

  As described above, the average value of the luminance value for each line is used as the image feature amount for each line in order to avoid the influence of the fluctuation of the luminance value due to noise or the like generated on the face image G as described above. It is. In other words, if there is noise or dirt on the original image G on the face image G to be searched, the luminance value of the pixel in that portion is greatly different from the luminance value of the neighboring pixels. If an average value of the values is calculated and used as an image feature amount for each line, it is possible to avoid the inconvenience of erroneously detecting that portion as the top of the head.

  Next, when the average luminance value (image feature amount) for each line in the crown detection window W is calculated in this way, the process proceeds to the next step S107, and the average luminance value for each line is further determined. By smoothing in the vertical direction (Y direction in the figure), abrupt changes in the image feature amount due to noise on the image, contamination of the original image, etc. are avoided as described above. Although this smoothing method is not particularly limited, for example, it is appropriate to use an average luminance value of other lines of about 1 to 3 lines above and below contacting the vertical direction with the target line as the center.

When the average luminance value for each line is calculated in this way, the process proceeds to step S109, and the calculated average luminance value for each line is vertically below the upper end in the top detection window W ( The amount of change is obtained by plotting sequentially in the direction (Y direction), and the position where the amount of change is equal to or greater than the threshold value is regarded as the top of the head.
For example, as shown in FIG. 8, the image feature amount composed of the average luminance value calculated for each line is plotted in the vertical direction (Y direction) in the top detection window W for each line, and the amount of change is plotted. A position where is more than a threshold value (for example, 10% of the average value of the image feature amount) from the initial average value is regarded as the top of the head.

The initial average value here is an average value of image feature amounts of lines near the upper base of the top detection window (for example, about 6 lines).
In the case of FIG. 8, since the image feature amount exceeds the threshold value in the vicinity of Y ₁ , the position of Y ₁ can be regarded as the top of the head, the range between Y ₀ and Y ₁ can be regarded as the background, and the position above Y ₁ can be regarded as the head. it can.
Note that the position that exceeds the threshold value in this way is regarded as the top of the head, as described above, even with a uniform background due to noise or dirt generated in the face image G, there is a slight difference in the image feature amount. This is to avoid erroneous detection due to variations, but in the case of a face image G or the like in which variations in image feature amount due to noise, dirt, etc. hardly occur, this threshold value may be small. On the contrary, in the case of the face image G covered with noise or dirt, the threshold value may be even larger.

  In the case of FIG. 8, an initial average value of image feature values is set and a threshold value is set based on the initial average value. However, as shown in FIG. And the distance between them may be set as a threshold value. In the example of FIG. 9, when the image feature amount at the position Y2 is compared with the image feature amount at the position Y3, the distance exceeds the threshold value, so the vicinity of Y3 is regarded as the top of the head.

If the top of the head is detected in this way, as shown in FIGS. 7 (a) and 7 (b), the human face is positioned so that the position of the top of the head becomes the same height as the standard top of the head. Move the whole thing.
In FIG. 7 (a), since the top of the human face is at a very low position, as shown in FIG. 7 (b), the human face is moved vertically upward to match the top of the head with the standard position. be able to. In FIG. 7 (a) and the like, the image below the person's neck is cut off, but in reality, the image of the hidden portion also exists as it is, and the person's face as shown in FIG. 7 (b). As a result of the movement, there is no inconvenience that a blank portion is generated in the display frame Y (lower part).

On the other hand, in the above-described embodiment, the average luminance value for each line is adopted as the image feature amount, but as described above, depending on the color of the hair, the lighting conditions, etc., the luminance value between the background and the top of the head is used. A large change does not appear, and the change amount of the image feature amount may not exceed the threshold value near the top of the head.
For this reason, when there is no significant change in the luminance value as described above, the luminance value of the pixel is changed as the image feature amount, and if the hue angle is used, the image feature amount is significantly changed, and the person is Can be accurately detected.

  FIG. 10 shows the top detection flow when the hue angle is used as the image feature amount as described above, and the first steps S101 and S103 detect the face in the same manner as when the luminance value is adopted. After setting the top detection window at the top of the head, the horizontal average value of the hue angle of the detection window is obtained (step S111), and the average hue angle is smoothed in the vertical direction within the detection window (step S111). S113) Finally, the amount of change is obtained, and if the amount of change is equal to or greater than the threshold value, the position is regarded as the top of the head (step S115).

As a result, even when the top of the person cannot be detected with the image feature amount based on the luminance value, the top of the person can be accurately detected.
Note that “luminance value” and “hue angle” in the present invention have the same meaning as generally defined in the field of image processing.
“Hue” is an attribute that indicates the difference in color. “Hue angle” is a color index diagram such as “Munsell's hue ring” or “Munsell's color solid”. Is the angle to the position of the hair color. For example, according to “Munsell's hue ring”, when the reference background color is “Blue”, “Yellow” or “Red” has a larger hue angle than “Green” (image) The feature amount is large).

FIG. 11 shows a processing flow of the image feature quantity selection unit 26 and the head top detection unit 34 described above, and considers the case where the top of the person cannot be detected by the image feature quantity based on the luminance value. An example of a flow for detecting the top of the head is shown.
As in the flow of FIG. 3, this head top detection flow first detects a human face included in an image, sets a head top detection window W at the head top of the human face, After obtaining the average value of the luminance values in the horizontal direction, the average luminance value is smoothed in the vertical direction and then the amount of change in the average luminance value is obtained (steps S201 to S209).

  Then, in the next determination step S211, it is determined whether or not the amount of change has exceeded the threshold value. When it is determined that the amount has exceeded (yes), the process proceeds to step S213 where the amount of change in the average luminance value exceeds the threshold value. Is determined to be the top of the head, but the process ends, but when it is determined that the amount of change does not exceed the threshold (No), the process proceeds to step S215 and this time based on the hue angle as shown in FIG. A change amount of the image feature amount is obtained (steps S215 to S219). Then, when it is determined that the amount of change has exceeded the threshold value (yes) after that, the process proceeds to step S223 and the position where the amount of change in the average luminance value exceeds the threshold value is regarded as the top of the head. When the process ends, but when it is determined that the amount of change does not exceed the threshold value (No), the process proceeds to step S225 to notify the user of this system that “the top of the head cannot be detected” ( Display) and the process is terminated.

  Here, as a case to reach step S225, in short, when both the luminance value and the hue angle of the hair portion (the top region) show substantially the same value as that of the background, for example, the color of the hair against the blue background Is blue and the saturation is almost the same, but detection of the top of the head in such a case seems to be difficult even by human vision with excellent sensitivity, and at the time of shooting etc. This is an extremely rare case where the photographer would consciously avoid the occurrence of such a case.

Therefore, the top of the head can be reliably detected by employing either or both of the average luminance and the average hue angle as described above.
In FIG. 11, the image feature quantity based on the luminance value is first obtained, and then the image feature quantity based on the hue angle is obtained. However, after the image feature quantity based on the hue angle is obtained first, the image based on the luminance value is obtained. The feature amount may be obtained, and the image feature amount of both may be used at the same time.

  Incidentally, since the change in luminance is easily recognized visually, it is frequently used as an image feature amount in other image processing techniques.

It is a block diagram which shows one Embodiment of a head part detection system. It is a block diagram which shows the hardware which comprises a head part detection system. It is a flowchart figure which shows one Embodiment of the top part detection method. It is a figure which shows an example of the face image used as a head-top part detection target. It is a figure which shows the state which set the face detection frame to the face image. It is a figure which shows the state which set the top part detection window in the upper part of the face detection frame. It is a figure which shows the state which detects a head top part and corrects the position. It is a figure which shows the relationship between an image feature-value and its variation | change_quantity. It is a figure which shows the relationship between an image feature-value and its variation | change_quantity. It is a flowchart figure which shows other embodiment of the top part detection method. It is a flowchart figure which shows other embodiment of the top part detection method.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 10 ... Image reading means, 12 ... Face detection means, 14 ... Parietal part detection window setting means, 16 ... Image feature-value calculation means, 18 ... Parietal part detection means, 22 ... Luminance calculation part, 24 ... Hue angle calculation part, 26 ... Image feature amount selection unit, 32 ... Threshold setting unit, average value calculation unit, 100 ... Head detection system, 40 ... CPU, 41 ... RAM, 42 ... ROM, 43 ... Auxiliary storage device, 44 ... Output device, 45 ... Input device 46 ... Input / output interface (I / F), 47 ... Bus, F ... Face detection frame, G ... Face image, M ... Marker, Y ... Display frame, W ... Top detection window.

Claims (12)

A method for detecting the top of a human face from an image including a human face,
After detecting a face image in a range not including the top of the person's face, and setting a top detection window having a size including the top of the person's face above the detected face image,
A method for detecting the top of the head, wherein an image feature amount in the top portion detection window is calculated in the vertical direction, and a position where the image feature amount changes to a threshold value or more is set as the top portion of the human face. The method of detecting the top of the head according to claim 1,
A method of detecting the top of the head, characterized in that the magnitude of luminance is used as the image feature amount. The method of detecting the top of the head according to claim 1,
A method for detecting the top of a head, wherein a magnitude of a hue angle is used as the image feature amount. In the top detection method according to any one of claims 1 to 3,
The method of detecting the top of the head, wherein the image feature amount uses an average value in a horizontal direction within the top of the head detection window. In the method for detecting the top of the head in the face image according to any one of claims 1 to 4,
The method for detecting the top of the head, wherein the image feature amount is automatically switched according to an image using either or both of luminance and hue angle. The method for detecting the top of the head according to claim 1,
A method of detecting the top of the head, wherein the top of the head detection window is rectangular and has a width and height wider than the face width of the human face. A system for detecting the top of a human face from an image including a human face,
Image reading means for reading an image including the human face;
A face detection means for detecting a face image in a range not including the top of the human face from the image read by the image reading means, and setting a face detection frame in the detected range;
A top detection window setting means for setting a top detection window having a size including the top of the human face at the top of the face detection frame;
Image feature amount calculating means for calculating the image feature amount in the top detection window in the vertical direction;
A vertex detection system comprising: a vertex detection unit that uses the position where the image feature value calculated by the image feature value calculation unit changes to a threshold value or more as a vertex of the human face. The head detection system for a human face according to claim 7,
The image feature amount calculating means includes:
A luminance calculation unit for calculating the luminance of each pixel as an image feature amount;
A hue angle calculation unit that calculates the hue angle of each pixel as an image feature amount;
An image feature amount selection unit that selects any one of these image feature amounts; In the head detection system for a human face according to claim 7 or 8,
The crown detection means is
A threshold value setting unit for setting a threshold value of the image feature amount;
A top part detection system comprising either or both of an average value calculation part for obtaining an average of horizontal image feature amounts in the top part detection window. A program for detecting the top of a person's face from an image including a person's face,
Image reading means for reading an image including the human face;
A face detection means for detecting a face image in a range not including the top of the human face from the image read by the image reading means, and setting a face detection frame in the detected range;
A top detection window setting means for setting a top detection window having a size including the top of the human face at the top of the face detection frame;
Image feature amount calculating means for calculating the image feature amount in the top detection window in the vertical direction;
A parietal part detection program for causing a computer to realize a parietal part detection unit that uses the position where the image feature quantity calculated by the image feature quantity calculation unit changes to a threshold value or more as the parietal part of the human face. In the human face head top detection program according to claim 10,
The image feature amount calculating means includes:
A luminance calculation function for calculating the luminance of each pixel as an image feature amount;
A hue angle calculation function for calculating the hue angle of each pixel as an image feature amount;
An image feature amount selection function for selecting any one of these image feature amounts; In the human face head top detection program according to claim 10 or 11,
The crown detection means is
A threshold value setting unit for setting a threshold value of the image feature amount;
A top part detection program comprising one or both of an average value calculation part for obtaining an average of horizontal image feature amounts in the top part detection window.

Images